interscript 0.1.0 → 0.1.5

data/maps/alalc-mkd-cyrl-latn-1997.yaml

@@ -0,0 +1,114 @@
+---
+authority_id: alalc
+id: 1997
+language: mkd
+source_script: Cyrl
+destination_script: Latn
+name: Serbian and Makedonian Romanization, ALA-LC 1997 System
+url: http://catdir.loc.gov/catdir/cpso/romanization/serbian.pdf
+creation_date: 1997
+description: ALA-LC Romanization table for Serbian and Makedonian.
+
+notes:
+  - Special characters in romanization
+    Đ - D with crossbar (upper case). USMARC hexadecimal code A3.
+    đ - d with crossbar (upper case). USMARC hexadecimal code B3.
+
+  - Character modifiers in romanization
+    ´ - acute. USMARC hexadecimal code E2.
+    ˇ - hachek. USMARC hexadecimal code E9.
+
+tests:
+  - source: Општина Ердут
+    expected: Opština Erdut
+  - source: Општина Двор
+    expected: Opština Dvor
+  - source: ЛУЃЕ луѓе
+    expected: LUǴE luǵe
+  - source: ЅВЕЗДА ѕвезда Ѕвезда
+    expected: DZVEZDA dzvezda Dzvezda
+  - source: ЌАРУВАЊЕ ќарување
+    expected: ḰARUVANJE ḱaruvanje
+
+map:
+  postrules:
+    # DZ
+    - pattern: "((?<=[[:upper:]])Dz(?=[[:upper:]])?|(?<=[[:upper:]])?Dz(?=[[:upper:]]))"
+      result: "DZ"
+    #LJ
+    - pattern: "((?<=[[:upper:]])Lj(?=[[:upper:]])?|(?<=[[:upper:]])?Lj(?=[[:upper:]]))"
+      result: "LJ"
+    #NJ
+    - pattern: "((?<=[[:upper:]])Nj(?=[[:upper:]])?|(?<=[[:upper:]])?Nj(?=[[:upper:]]))"
+      result: "NJ"
+    #DŽ
+    - pattern: "((?<=[[:upper:]])Dž(?=[[:upper:]])?|(?<=[[:upper:]])?Dž(?=[[:upper:]]))"
+      result: "DŽ"
+
+  characters:
+    "\u0410": "A"
+    "\u0411": "B"
+    "\u0412": "V"
+    "\u0413": "G"
+    "\u0403": "\u01F4" # Ǵ
+    "\u0414": "D"
+    "\u0402": "\u0110" # Đ
+    "\u0415": "E"
+    "\u0416": "\u005a\u030c" # Ž
+    "\u0417": "Z"
+    "\u0405": "Dz"
+    "\u0418": "I"
+    "\u0408": "J"
+    "\u041A": "K"
+    "\u040C": "\u1E30" # Ḱ
+    "\u041B": "L"
+    "\u0409": "Lj"
+    "\u041C": "M"
+    "\u041D": "N"
+    "\u040A": "Nj"
+    "\u041E": "O"
+    "\u041F": "P"
+    "\u0420": "R"
+    "\u0421": "S"
+    "\u0422": "T"
+    "\u040B": "\u0043\u0301" # Ć
+    "\u0423": "U"
+    "\u0424": "F"
+    "\u0425": "H"
+    "\u0426": "C"
+    "\u0427": "\u0043\u030c" # Č
+    "\u040F": "D\u007a\u030c" # Dž
+    "\u0428": "\u0053\u030c" # Š
+    "\u0430": "a"
+    "\u0431": "b"
+    "\u0432": "v"
+    "\u0433": "g"
+    "\u0453": "\u01F5" # ǵ
+    "\u0434": "d"
+    "\u0452": "\u0111" # đ
+    "\u0435": "e"
+    "\u0436": "\u007a\u030c" # ž
+    "\u0437": "z"
+    "\u0455": "dz"
+    "\u0438": "i"
+    "\u0458": "j"
+    "\u043A": "k"
+    "\u045C": "\u1E31" # ḱ
+    "\u043B": "l"
+    "\u0459": "lj"
+    "\u043C": "m"
+    "\u043D": "n"
+    "\u045A": "nj"
+    "\u043E": "o"
+    "\u043F": "p"
+    "\u0440": "r"
+    "\u0441": "s"
+    "\u0442": "t"
+    "\u045B": "\u0063\u0301" # ć
+    "\u0443": "u"
+    "\u0444": "f"
+    "\u0445": "h"
+    "\u0446": "c"
+    "\u0447": "\u0063\u030c" # č
+    "\u045F": "d\u007a\u030c" # dž
+    "\u0448": "\u0073\u030c" # š

data/maps/alalc-rus-Cyrl-Latn-1997.yaml

@@ -0,0 +1,222 @@
+---
+authority_id: alalc
+id: 1997
+language: rus
+source_script: Cyrl
+destination_script: Latn
+name: ALA-LC Romanization System 1997
+url: https://en.wikipedia.org/wiki/ALA-LC_romanization_for_Russian
+creation_date: 1997
+description: |
+  The ALA-LC Romanization tables for Slavic alphabets is a set of standards for romanization of texts
+  in various writing systems used in North American libraries and publications.
+  This version was published by the American Library Association and the Library of Congress in 1997.
+  This article is about the ALA-LC method of transliteration of Russian-language text from Cyrillic script to Latin script.
+
+  The formal, unambiguous version of the system requires some diacritics and two-letter tie characters which are often omitted in practice.
+
+notes:
+  - Pre-1918 letters skipped
+  - Pre-18th century letters skipped
+
+tests:
+  - source: Азов
+    expected: Azov
+  - source: Тамбов
+    expected: Tambov
+  - source: Барнаул
+    expected: Barnaul
+  - source: Кубань
+    expected: Kubanʹ
+  - source: Владимир
+    expected: Vladimir
+  - source: Ульяновск
+    expected: Ulʹi͡anovsk
+  - source: Грозный
+    expected: Groznyǐ
+  - source: Волгодонск
+    expected: Volgodonsk
+  - source: Дзержинский
+    expected: Dzerzhinskiǐ
+  - source: Нелидово
+    expected: Nelidovo
+  - source: Елизово
+    expected: Elizovo
+  - source: Чебоксары
+    expected: Cheboksary
+  - source: Ёлкин
+    expected: Ëlkin
+  - source: Озёрный 
+    expected: Ozërnyǐ
+  - source: Жуков
+    expected: Zhukov
+  - source: Лужники
+    expected: Luzhniki
+  - source: Звенигород
+    expected: Zvenigorod
+  - source: Вязьма
+    expected: Vi͡azʹma
+  - source: Иркутск
+    expected: Irkutsk
+  - source: Апатиты
+    expected: Apatity
+  - source: Йошкар-Ола
+    expected: Ǐoshkar-Ola
+  - source: Бийск
+    expected: Biǐsk
+  - source: Киров
+    expected: Kirov
+  - source: Енисейск
+    expected: Eniseǐsk
+  - source: Ломоносов
+    expected: Lomonosov
+  - source: Нелидово
+    expected: Nelidovo
+  - source: Менделеев
+    expected: Mendeleev
+  - source: Каменка
+    expected: Kamenka
+  - source: Новосибирск
+    expected: Novosibirsk
+  - source: Кандалакша
+    expected: Kandalaksha
+  - source: Омск
+    expected: Omsk
+  - source: Красноярск
+    expected: Krasnoi͡arsk
+  - source: Петрозаводск
+    expected: Petrozavodsk
+  - source: Серпухов
+    expected: Serpukhov
+  - source: Ростов
+    expected: Rostov
+  - source: Северобайкальск
+    expected: Severobaǐkalʹsk
+  - source: Сковородино
+    expected: Skovorodino
+  - source: Чайковский
+    expected: Chaǐkovskiǐ
+  - source: Тамбов
+    expected: Tambov
+  - source: Мытищи
+    expected: Mytishchi
+  - source: Углич
+    expected: Uglich
+  - source: Дудинка
+    expected: Dudinka
+  - source: Фурманов
+    expected: Furmanov
+  - source: Уфа
+    expected: Ufa
+  - source: Хабаровск
+    expected: Khabarovsk
+  - source: Прохладный
+    expected: Prokhladnyǐ
+  - source: Цимлянск
+    expected: T͡Simli͡ansk
+  - source: Ельцин
+    expected: Elʹt͡sin
+  - source: Чебоксары
+    expected: Cheboksary
+  - source: Печора
+    expected: Pechora
+  - source: Шахтёрск
+    expected: Shakhtërsk
+  - source: Мышкин
+    expected: Myshkin
+  - source: Щёлково
+    expected: Shchëlkovo
+  - source: Ртищево
+    expected: Rtishchevo
+  - source: Подъездной
+    expected: Podʺezdnoǐ
+  - source: Ыттык-Кёль
+    expected: Yttyk-Këlʹ
+  - source: Тында
+    expected: Tynda
+  - source: Тюмень
+    expected: Ti͡umenʹ
+  - source: Электрогорск
+    expected: Ėlektrogorsk
+  - source: Радиоэлектроника
+    expected: Radioėlektronika
+  - source: Юбилейный
+    expected: I͡Ubileǐnyǐ
+  - source: Ключевская
+    expected: Kli͡uchevskai͡a
+  - source: Якутск
+    expected: I͡Akutsk
+  - source: Брянск
+    expected: Bri͡ansk
+
+map:
+  characters:
+    "\u0027": ""         # '
+    "\u0410": "A"        # А
+    "\u0411": "B"        # Б
+    "\u0412": "V"        # В
+    "\u0413": "G"        # Г
+    "\u0414": "D"        # Д
+    "\u0415": "E"        # Е
+    "\u0401": "Ë"        # Ё
+    "\u0416": "Zh"       # Ж
+    "\u0417": "Z"        # З
+    "\u0418": "I"        # И
+    "\u0419": "\u01CF"   # Й
+    "\u041A": "K"        # К
+    "\u041B": "L"        # Л
+    "\u041C": "M"        # М
+    "\u041D": "N"        # Н
+    "\u041E": "O"        # О
+    "\u041F": "P"        # П
+    "\u0420": "R"        # Р
+    "\u0421": "S"        # С
+    "\u0422": "T"        # Т
+    "\u0423": "U"        # У
+    "\u0424": "F"        # Ф
+    "\u0425": "Kh"       # Х
+    "\u0426": "T\u0361S" # Ц
+    "\u0427": "Ch"       # Ч
+    "\u0428": "Sh"       # Ш
+    "\u0429": "Shch"     # Щ
+    "\u042A": "ʺ"        # Ъ
+    "\u042B": "Y"        # Ы
+    "\u042C": "ʹ"        # Ь
+    "\u042D": "E\u0307"  # Э
+    "\u042E": "I\u0361U" # Ю
+    "\u042F": "I\u0361A" # Я
+
+    "\u0430": "a"        # а
+    "\u0431": "b"        # б
+    "\u0432": "v"        # в
+    "\u0433": "g"        # г
+    "\u0434": "d"        # д
+    "\u0435": "e"        # e
+    "\u0451": "ë"        # ё
+    "\u0436": "zh"       # ж
+    "\u0437": "z"        # з
+    "\u0438": "i"        # и
+    "\u0439": "\u01d0"   # й
+    "\u043A": "k"        # к
+    "\u043B": "l"        # л
+    "\u043C": "m"        # м
+    "\u043D": "n"        # н
+    "\u043E": "o"        # о
+    "\u043F": "p"        # п
+    "\u0440": "r"        # р
+    "\u0441": "s"        # с
+    "\u0442": "t"        # т
+    "\u0443": "u"        # у
+    "\u0444": "f"        # ф
+    "\u0445": "kh"       # х
+    "\u0446": "t\u0361s" # ц
+    "\u0447": "ch"       # ч
+    "\u0448": "sh"       # ш
+    "\u0449": "shch"     # щ
+    "\u044A": "ʺ"        # ъ
+    "\u044B": "y"        # ы
+    "\u044C": "ʹ"        # ь
+    "\u044D": "e\u0307"  # э
+    "\u044E": "i\u0361u" # ю
+    "\u044F": "i\u0361a" # я
+

data/maps/alalc-rus-Cyrl-Latn-2012.yaml

@@ -0,0 +1,162 @@
+---
+authority_id: alalc
+id: 2012
+language: rus
+source_script: Cyrl
+destination_script: Latn
+name: ALA-LC Romanization System 2012
+url: https://www.loc.gov/catdir/cpso/romanization/russian.pdf
+creation_date: 2012
+description: |
+  The ALA-LC Romanization tables for Slavic alphabets is a set of standards for romanization of texts
+  in various writing systems used in North American libraries and publications.
+  The latest version was published by the American Library Association and the Library of Congress in 2012.
+
+notes:
+  - The letters І, Ѣ, Ѳ and Ѵ were eliminated in the orthographic reform of 1918. For other obsolete letters appearing in Russian texts, consult the Church Slavic table. 
+
+tests:
+  - source: Азов
+    expected: Azov
+  - source: Тамбов
+    expected: Tambov
+  - source: Барнаул
+    expected: Barnaul
+  - source: Кубань
+    expected: Kubanʹ
+  - source: Владимир
+    expected: Vladimir
+  - source: Ульяновск
+    expected: Ulʹi͡anovsk
+  - source: Грозный
+    expected: Groznyǐ
+  - source: Волгодонск
+    expected: Volgodonsk
+  - source: Дзержинский
+    expected: Dzerzhinskiǐ
+  - source: Нелидово
+    expected: Nelidovo
+  - source: Елизово
+    expected: Elizovo
+  - source: Чебоксары
+    expected: Cheboksary
+  - source: Ёлкин
+    expected: Ëlkin
+  - source: Озёрный 
+    expected: Ozërnyǐ
+  - source: Жуков
+    expected: Zhukov
+  - source: Лужники
+    expected: Luzhniki
+  - source: Звенигород
+    expected: Zvenigorod
+  - source: Вязьма
+    expected: Vi͡azʹma
+  - source: Иркутск
+    expected: Irkutsk
+  - source: Апатиты
+    expected: Apatity
+  - source: Йошкар-Ола
+    expected: Ǐoshkar-Ola
+  - source: Бийск
+    expected: Biǐsk
+  - source: Киров
+    expected: Kirov
+  - source: Енисейск
+    expected: Eniseǐsk
+  - source: Ломоносов
+    expected: Lomonosov
+  - source: Нелидово
+    expected: Nelidovo
+  - source: Менделеев
+    expected: Mendeleev
+  - source: Каменка
+    expected: Kamenka
+  - source: Новосибирск
+    expected: Novosibirsk
+  - source: Кандалакша
+    expected: Kandalaksha
+  - source: Омск
+    expected: Omsk
+  - source: Красноярск
+    expected: Krasnoi͡arsk
+  - source: Петрозаводск
+    expected: Petrozavodsk
+  - source: Серпухов
+    expected: Serpukhov
+  - source: Ростов
+    expected: Rostov
+  - source: Северобайкальск
+    expected: Severobaǐkalʹsk
+  - source: Сковородино
+    expected: Skovorodino
+  - source: Чайковский
+    expected: Chaǐkovskiǐ
+  - source: Тамбов
+    expected: Tambov
+  - source: Мытищи
+    expected: Mytishchi
+  - source: Углич
+    expected: Uglich
+  - source: Дудинка
+    expected: Dudinka
+  - source: Фурманов
+    expected: Furmanov
+  - source: Уфа
+    expected: Ufa
+  - source: Хабаровск
+    expected: Khabarovsk
+  - source: Прохладный
+    expected: Prokhladnyǐ
+  - source: Цимлянск
+    expected: T͡Simli͡ansk
+  - source: Ельцин
+    expected: Elʹt͡sin
+  - source: Чебоксары
+    expected: Cheboksary
+  - source: Печора
+    expected: Pechora
+  - source: Шахтёрск
+    expected: Shakhtërsk
+  - source: Мышкин
+    expected: Myshkin
+  - source: Щёлково
+    expected: Shchëlkovo
+  - source: Ртищево
+    expected: Rtishchevo
+  - source: Подъездной
+    expected: Podʺezdnoǐ
+  - source: Ыттык-Кёль
+    expected: Yttyk-Këlʹ
+  - source: Тында
+    expected: Tynda
+  - source: Тюмень
+    expected: Ti͡umenʹ
+  - source: Электрогорск
+    expected: Ėlektrogorsk
+  - source: Радиоэлектроника
+    expected: Radioėlektronika
+  - source: Юбилейный
+    expected: I͡Ubileǐnyǐ
+  - source: Ключевская
+    expected: Kli͡uchevskai͡a
+  - source: Якутск
+    expected: I͡Akutsk
+  - source: Брянск
+    expected: Bri͡ansk
+
+map:
+  inherit: alalc-rus-Cyrl-Latn-1997
+
+  characters:
+    '\u0406': "\u012A"
+    '\u0456': "\u012B"
+
+    '\u0462': "I\u0361E" # Ѣ
+    '\u0463': "i\u0361e" # ѣ
+
+    '\u0472': "F\u0307" # Ѳ
+    '\u0473': "f\u0307" # ѳ
+
+    '\u0474': "Y\u0307" # Ѵ
+    '\u0475': "y\u0307" # ѵ